docs/READMEcode/code/runREADME.R

In tlcaputi/gtrendR: An R Complement to the gtrends Python package

# # Only run this once
# library(devtools)
# devtools::install_github("tlcaputi/gtrendR")

# Run this every time
pacman::p_load(grid)
library("gtrendR")
ROOTPATH <- "C:/Users/tcapu/Google Drive/modules/gtrendR/docs/READMEcode"
setwd(ROOTPATH)

code_files <- dir("C:/Users/tcapu/Google Drive/modules/gtrendR/R", full.name = T)
for(f in code_files) source(f)


US_df <- run_arima(
  df = read.csv("./input/handwashing_day.csv", header = T, stringsAsFactor = F), # Data from gtrends
  interrupt = "2020-03-01", # Interruption point in your data
  geo = "US", # geography you want to use
  kalman = T # If True, uses Kalman method to impute time series
)


panA <- line_plot(
  US_df, # data from run_arima
  geo = 'US', # geography you wnat to use

  ## Create a vertical "interruption" line in your plot
  interrupt = "2020-03-01", # Date of an interruption
  linelabel = "COVID19",

  ## Plot arguments
  beginplot = T, # Start date for the plot. If T, beginning of data
  endplot = T, # End date for the plot. If T, end of data
  title = NULL, # If NULL, no Title
  xlab = "Date", # x axis label
  lbreak = "3 month", # Space between x-axis tick marks
  xfmt = date_format("%b %Y"), # Format of dates on x axis
  ylab = "Query Fraction\n(Per 10 Million Searches)", # y axis label
  lwd = 0.3, # Width of the line

  ## Set a colorscheme
  colorscheme = "blue",  # Color schemes set in this package "red", 'blue" or "jamaim"

  # ... customize any color using these
  hicol = NA, # Searches line color
  opcol = NA, # Color of point on top of spike


  ## Saving arguments
  save = T, # If T, save plot
  outfn = './output/panA.png', # Location to save plot
  width = 6, # Width in inches
  height = 3 # Height in inches
)



panB <- arima_plot(
  US_df, ## data from run_arima

  ## Create a vertical "interruption" line in your plot
  interrupt = "2020-03-01", # Date of an interruption
  linelabel = "",
  linelabelpos = 0.02, # Where the label goes near the interruption line


  ## Plot Arguments
  beginplot = "2019-09-01", # Start date for the plot. If T, beginning of data
  endplot = "2020-04-01", # End date for the plot. If T, end of data
  title = NULL, # If NULL, no Title
  xlab = "Date", # x axis label
  lbreak = "1 month", # Space between x-axis tick marks
  xfmt = date_format("%b %Y"), # Format of dates on x axis
  ylab = "Query Fraction\n(Per 10 Million Searches)", # y axis label
  lwd = 1, # Width of the line
  label = T, # put increase in searches in plot
  labsize = 0.8, # size of label

  ## Set a colorscheme
  colorscheme = "blue",  # Color schemes set in this package "red", 'blue" or "jamaim"

  # ... customize any color using these
  hicol = NA, # Actual line color
  locol = NA, # Expected line color
  nucol = NA, # Excess polygon color


  ## Saving arguments
  save = T, # If T, save plot
  outfn = './output/panB.png', # Location to save plot
  width = 6, # Width in inches
  height = 3 # Height in inches

)



panC <- arima_ciplot(
  US_df, ## data from run_arima

  ## Create a vertical "interruption" line in your plot
  interrupt = "2020-03-01", # Date of an interruption

  ## Plot Arguments
  beginplot = T, # Start date for the plot. If T, beginning of data
  endplot = "2020-04-01", # End date for the plot. If T, end of data
  title = NULL, # If NULL, no Title
  xlab = "Date", # x axis label
  lbreak = "1 week", # Space between x-axis tick marks
  xfmt = date_format("%b-%d"), # Format of dates on x axis
  ylab = "Greater than Expected (%)", # y axis label
  lwd = 1, # Width of the line

  ## Set a colorscheme
  colorscheme = "blue",  # Color schemes set in this package "red", 'blue" or "jamaim"

  # ... customize any color using these
  hicol = NA, # Actual color
  nucol = NA, # Polygon color


  ## Saving arguments
  save = T, # If T, save plot
  outfn = './output/panC.png', # Location to save plot
  width = 6, # Width in inches
  height = 3 # Height in inches

)

panC <- panC +
  scale_x_date(
    limits = c(ymd("2020-03-01"), ymd("2020-04-01")),
    date_breaks = "1 day",
    labels = function(x) ifelse(as.numeric(x) %% 2 != 0, "", format(x, format = "%b %d"))
  ) +
  theme(axis.text.x = element_text(angle = 45, vjust = 1.0, hjust = 1.0))

title <- ggdraw() +
  draw_label(
    "Google Searches for Hand Washing",
    fontface = 'bold',
    hjust = 0.5
  ) +
  theme(
    plot.margin = margin(0, 0, 0, 7)
  )

fig <- plot_grid(panA, panB, panC, labels=c(LETTERS[1:3]), ncol=1, nrow=3, rel_heights=c(1,1,1))
fig <- plot_grid(title, fig, ncol = 1, rel_heights = c(0.1, 1))
save_plot("./output/Fig1.png", fig, base_width=6, base_height=10)








out <- state_pct_change(
  df = read.csv("./input/handwashing_day.csv", header = T, stringsAsFactor = F), ## Data from gtrends

  ## You will need to decide on the timeframes for "before" and "after"
  beginperiod = NA, # If not NA, this is the start of the "before" period
  preperiod = 90, # If beginperiod is NA, this uses 90 days before the interruption
  interrupt = "2020-03-01", # The date of the interruption
  endperiod = "2020-04-01", # The after period is the interruption to the endperiod

  ## Scale Legend
  scaletitle = "% Increase\nin Searches",
  scalelimits = NULL, # Vector of length 2 with lower and upper limit

  ## Set a colorscheme
  colorscheme = "blue",  # Color schemes set in this package "red", 'blue" or "jamaim"

  # ... customize any color using these
  highcol = NA, # Color for highest percent change
  midcol = NA, # Color for 0 percent change
  lowcol = NA, # Color for lowest percent change
  linecol = "gray", # Line between states

  ## Saving arguments
  save = T, # If T, save plot
  outfn = './output/panD.png', # Location to save plot
  width = 6, # Width in inches
  height = 3, # Height in inches

  ## Get data back from this function
  return_df = T,
  # If this is True...
  bootstrap = T, ## Bootstrap confidence intervals for pct change
  bootnum = 1000, # Number of bootstraps
  alpha = 0.05 # Alpha value for CIs

)

panD <- out[[2]]


state_list <- state_arima(
  data = read.csv("./input/handwashing_day.csv", header = T, stringsAsFactor = F), ## Data from gtrends
  interrupt = "2020-03-01", ## Interruption point
  begin = T, ## Beginning of the time period to use
  end = T, ## End of the time period to use
  kalman = T ## If True, Kalman impute NAs in the time series
)




panE <- state_arima_spaghetti(
  state_list, # data from state_arima
  interrupt = "2020-03-01", # should be the same as state_arima

  ## Plot Arguments
  beginplot = "2020-03-01", # Start date for the plot. If T, beginning of data
  endplot = "2020-04-03", # End date for the plot. If T, end of data
  title = NULL, # If NULL, no Title
  xlab = "Date", # x axis label
  lbreak = "1 week", # Space between x-axis tick marks
  xfmt = date_format("%b-%d"), # Format of dates on x axis
  ylab = "Query Fraction\n(Per 10 Million Searches)", # y axis label
  lwd = 1, # Width of the line
  ylim = c(NA, NA), # y axis limts

  ## Spaghetti specific adjustments
  spaghettialpha = 0.25, # How transparent do you want the spaghetti lines
  states_with_labels = c("US"), ## Add labels to the end of these
  states_to_exclude = c("IA"), ## Don't include these states
  right_margin = 0.05, # margin on right for labels

  ## Set a colorscheme
  colorscheme = "blue",  # Color schemes set in this package "red", 'blue" or "jamaim"

  # ... customize any color using these
  hicol = NA, # Color of US line
  locol = NA, # Color of other lines

  ## Saving arguments
  save = T, # If T, save plot
  outfn = './output/panE.png', # Location to save plot
  width = 6, # Width in inches
  height = 3 # Height in inches
)

panE <- panE + coord_cartesian(ylim = c(-10, 40))

panF <- state_arima_pctdiff(
  state_list, # data from state_arima

  ## Set a colorscheme
  colorscheme = "blue",  # Color schemes set in this package "red", 'blue" or "jamaim"

  # ... customize any color using these
  highcol = NA, # Color for highest percent change
  midcol = NA, # Color for 0 percent change
  lowcol = NA, # Color for lowest percent change
  linecol = "gray", # Line between states

  ## Scale Arguments
  scaletitle = "% Diff.\nin Searches",

  ## Saving arguments
  save = T, # If T, save plot
  outfn = './output/panF.png', # Location to save plot
  width = 6, # Width in inches
  height = 3 # Height in inches

)



title <- ggdraw() +
  draw_label(
    "Google Searches for Hand Washing",
    fontface = 'bold',
    hjust = 0.5
  ) +
  theme(
    plot.margin = margin(0, 0, 0, 7)
  )
fig <- plot_grid(panD, panE, panF, labels=c(LETTERS[4:6]), ncol=1, nrow=3, rel_heights=c(1.1, 1, 1.1))
fig <- plot_grid(title, fig, ncol = 1, rel_heights = c(0.05, 1))
save_plot("./output/Fig2.png", fig, base_width=7, base_height=12)




multiterms <- multi_term_arima(

  ## A folder containing all of your gtrends data
  input_dir = "./input",

  ## Which data to use
  geo = "US", # Geography you want to use
  terms_to_use = NA, # Terms you'd like to analyze. If NA then all terms
  timeframe_to_use = NA, # Only analyze data with filenames that contain a certain timeframe. If NA then all timeframes


  ## Parameters of time periods
  beginperiod = T, # Beginning of the before period, if T then beginning of data
  preperiod = 90, # If beginperiod is logical, preperiod is the number of days before interrupt to include in before period
  endperiod = T, # End of the end period, if T then end of data
  interrupt = "2020-03-01", # Date for interruption, splitting before and after periods


  ## Analytical arguments
  bootstrap = T, # Bootstrap CIs
  bootnum = 1000, # Number of bootstraps
  kalman = T # If T, impute with Kalman
)


panG <- multiterm_barplot(
  multiterm_list = multiterms,

  ## Graphing Parameters
  title = NULL, # If NULL, no Title
  xlab = "Terms", # x axis label
  label_df = NA, # Use a two-column dataframe to label the barplot x axis
  ylab = "Greater than Expected (%)", # y axis label
  space = 0.8, # space between bars
  barlabels = F,
  

  ## Set a colorscheme
  colorscheme = "blue",  # Color schemes set in this package "red", 'blue" or "jamaim"

  # ... customize any color using these
  hicol = NA, # Color of bars

  ## Saving arguments
  save = T, # If T, save plot
  outfn = './output/panG.png', # Location to save plot
  width = 6, # Width in inches
  height = 3 # Height in inches
)



panH <- multiterm_spaghetti(
  multiterm_list = multiterms,
  interrupt = "2020-03-01",
  terms_to_use = NA,

  ## Plot Arguments
  beginplot = "2020-01-01", # Start date for the plot. If T, beginning of data
  endplot = "2020-04-03", # End date for the plot. If T, end of data
  title = NULL, # If NULL, no Title
  xlab = "Date", # x axis label
  lbreak = "1 month", # Space between x-axis tick marks
  xfmt = date_format("%b-%Y"), # Format of dates on x axis
  ylab = "Query Fraction\n(Per 10 Million Searches)", # y axis label
  lwd = 1, # Width of the line
  ylim = c(NA, NA), # y axis limts

  ## Spaghetti specific adjustments
  spaghettialpha = 0.6, # How transparent do you want the spaghetti lines

  ## Set a colorscheme
  colorscheme = "blue",  # Color schemes set in this package "red", 'blue" or "jamaim"

  # ... customize any color using these
  hicol = NA, # Color of US line
  locol = NA, # Color of other lines

  ## Saving arguments
  save = T, # If T, save plot
  outfn = './output/panH.png', # Location to save plot
  width = 6, # Width in inches
  height = 3 # Height in inches
  )